HPC

Dell crafts mother of all graphics cards

16 GPUs jammed in 3U chassis


Here's just what you need to play Crysis. The bespoke server division of Dell, called Data Center Solutions and accounting for a sizeable portion of Dell's quarterly server volumes, has cooked up a PCI-Express expansion chassis that can house up to 16 GPU co-processors.

The new expansion chassis is part of the PowerEdge C class of machines that Dell launched in March. These PowerEdge C machines fall somewhere between the general purpose PowerEdge boxes Dell sells directly and through channel partners and the truly custom boxes made by DCS; they are not sold directly over the web, but rather through a sales engagement.

Generally speaking, the PowerEdge C machines are aimed at hyperscale computing jobs and focus on density and power efficiency. The C1100 is a 1U box and the C2100 is a 2U box, both with two sockets designed to run Intel's latest six-core Xeon 5600 processors; the C6100 puts four half-width two-socket Xeon 5600 systems into a 2U rack chassis, and is the densest rack server that Dell sells - with the possible exception of custom boxes no one but the customer and Dell know about.

The PowerEdge C410x is a PCI-Express expansion chassis which is meant to link to servers over PCI-Express links. According to Joe Sekel, a systems architect for the DCS unit at Dell, the 3U chassis comes with fan-outs and PCI-Express switches that let from one to four GPUs in the chassis to be allocated to a single server. Each GPU co-processor is wrapped in a metal skin that Sekel called a "taco," which directs the airflow around the GPU cards so they don't cause each other to overheat.

The C410x chassis puts ten GPUs in the front, where the cold aisle is in a data center, and six more in the back, where the hot aisle is. There is room for eight PCI-Express links and four power supplies (with 3+1 redundancy). The unit has eight fans that are all hot pluggable (7+1 redundancy), and the GPU co-processors themselves are also hot-pluggable once they are encased in the taco enclosure.

Basically, this is a GPU blade server, as you can see below:

Dell PowerEdge C410x GPU Chassis

Dell's PowerEdge C410x GPU enclosure.

The PowerEdge C410x chassis is certified to use the older Nvidia Tesla M1060 fanless GPU co-processors or the new Tesla M2050 fanless units, which debuted in May using the latest Fermi GPUs.

The fanless design of the M1060 and M2050 uses big ole heat sinks to take the heat off the GPU and the cooling fans of the server - or in this case, PCI expansion chassis - to keep the units from melting spectacularly. Dell has not certified the fanless FireStream GPU co-processors from AMD, launched in June, for the C410x enclosures.

The M2050 GPU co-processor is rated at the same 515 gigaflops of double-precision and 1.03 teraflops single-precision floating point performance, which means you can pack up to 16.48 teraflops of single-precision oomph into a single C410x chassis. The M2070 will have the same flops ratings, but will include 6GB of GDDR5 memory to the M2050's 3GB; the M2070 is not yet certified for the C410x because it is not yet shipping from Nvidia.

The AMD FireStream 9350 fanless GPU co-processor is rated at two teraflops SP and 400 gigaflops DP, and is something some customers probably want. The FireStream 9370 is a double-wide card rated at 2.64 teraflops SP and 528 gigaflops DP, but has 4GB of GDDR5 memory - twice that of the FireStream 9350.

At the moment, Dell is pitching the C410x as a companion to the C6100 four-server rack mounted machine, which would allow four GPUs to be linked to four servers, or one per Xeon CPU socket. Sekel says that the reason Dell created the C410x is that an oil and gas customer doing seismic processing wasn't sure if it needed two, three, or four GPUs per physical server to run its workloads. So Dell built a box that could dynamically allocate GPUs to machines.

The C410x is not restricted to GPU co-processors. Any PCI-Express 2.0 x16 device can, in theory, be wrapped in the metal taco, made hot-pluggable, and slid into the chassis if customers need it. Sekel said one obvious candidate was a Fusion-io PCI-Express flash drive, or any number of PCI-Express peripherals.

The only requirement is that each PCI-Express module has to burn 225 watts or less. The C410x uses 1,400 watt power supplies and has a maximum draw of 3,600 watts from the PCI-Express slots. That's a pretty tight power budget, with only 600 watts of headroom to spare for inefficiency, switching, fans, and der blinken lights.

The PowerEdge C410x chassis is available now. Dell does not give out pricing for any of the PowerEdge C class iron. ®

Similar topics


Other stories you might like

  • Share your experience: How does your organization introduce new systems?

    The answer is rarely obvious. Take part in our short poll and we'll find out together

    Reg Reader Survey The introduction of new systems into an organization is essential. If we stay still, if we continue to rely on legacy systems, if we fail to innovate – well, we (or, in reality, the company) will die. As business guru Sir John Harvey-Jones once put it: “If you are doing things the same way as two years ago, you are almost certainly doing them wrong.”

    But who should lead innovation in our companies? Who should be introducing new systems? The answer is not obvious.

    On one hand, the introduction of new systems into the business should be led by the business. In principle, the people doing the work, dealing with the suppliers, selling to the customers, are best placed to be standing up and saying: “We need the system to do X,” whether their motivation be to reduce cost, increase revenues, make products more efficiently, or even bolster our environmental credentials.

    Continue reading
  • These Rapoo webcams won't blow your mind, but they also won't break the bank

    And they're almost certainly better than a laptop jowel-cam

    Review It has been a long 20 months since Lockdown 1.0, and despite the best efforts of Google and Zoom et al to filter out the worst effects of built-in laptop webcams, a replacement might be in order for the long haul ahead.

    With this in mind, El Reg's intrepid reviews desk looked at a pair of inexpensive Rapoo webcams in search for an alternative to the horror of our Dell XPS nose-cam.

    Rapoo sent us its higher-end XW2K, a 2K 30fps device and, at the other end of the scale, the 720p XW170. Neither will break the bank, coming in at around £40 and £25 respectively from online retailers, but do include some handy features, such as autofocus and a noise cancelling microphone.

    Continue reading
  • It's one thing to have the world in your hands – what are you going to do with it?

    Google won the patent battle against ART+COM, but we were left with little more than a toy

    Column I used to think technology could change the world. Google's vision is different: it just wants you to sort of play with the world. That's fun, but it's not as powerful as it could be.

    Despite the fact that it often gives me a stomach-churning sense of motion sickness, I've been spending quite a bit of time lately fully immersed in Google Earth VR. Pop down inside a major city centre – Sydney, San Francisco or London – and the intense data-gathering work performed by Google's global fleet of scanning vehicles shows up in eye-popping detail.

    Buildings are rendered photorealistically, using the mathematics of photogrammetry to extrude three-dimensional solids from multiple two-dimensional images. Trees resolve across successive passes from childlike lollipops into complex textured forms. Yet what should feel absolutely real seems exactly the opposite – leaving me cold, as though I've stumbled onto a global-scale miniature train set, built by someone with too much time on their hands. What good is it, really?

    Continue reading
  • Why Cloud First should not have to mean Cloud Everywhere

    HPE urges 'consciously hybrid' strategy for UK public sector

    Sponsored In 2013, the UK government heralded Cloud First, a ground-breaking strategy to drive cloud adoption across the public sector. Eight years on, and much of UK public sector IT still runs on-premises - and all too often - on obsolete technologies.

    Today the government‘s message boils down to “cloud first, if you can” - perhaps in recognition that modernising complex legacy systems is hard. But in the private sector today, enterprises are typically mixing and matching cloud and on-premises infrastructure, according to the best business fit for their needs.

    The UK government should also adopt a “consciously hybrid” approach, according to HPE, The global technology company is calling for the entire IT industry to step up so that the public sector can modernise where needed and keep up with innovation: “We’re calling for a collective IT industry response to the problem,” says Russell MacDonald, HPE strategic advisor to the public sector.

    Continue reading
  • A Raspberry Pi HAT for the Lego Technic fan

    Sneaking in programming under the guise of plastic bricks

    There is good news for the intersection of Lego and Raspberry Pi fans today, as a new HAT (the delightfully named Hardware Attached on Top) will be unveiled for the diminutive computer to control Technic motors and sensors.

    Continue reading
  • Reg scribe spends week being watched by government Bluetooth wristband, emerges to more surveillance

    Home quarantine week was the price for an overseas trip, ongoing observation is the price of COVID-19

    Feature My family and I recently returned to Singapore after an overseas trip that, for the first time in over a year, did not require the ordeal of two weeks of quarantine in a hotel room.

    Instead, returning travelers are required to stay at home, wear a government-issued tracking device, and stay within range of a government-issued Bluetooth beacon at all times for a week … or else. No visitors are allowed and only a medical emergency is a ticket out. But that sounded easy compared to the hotel quarantine we endured in 2020.

    Continue reading
  • Intel teases 'software-defined silicon' with Linux kernel contribution – and won't say why

    It might enable activation of entirely new features on existing Xeon CPUs … or, you know, not

    Intel has teased a new tech it calls "Software Defined Silicon" (SDSi) but is saying almost nothing about it – and has told The Register it could amount to nothing.

    SDSi popped up around three weeks ago in a post to the Linux Kernel mailing list, in which an Intel Linux software engineer named David Box described it as "a post-manufacturing mechanism for activating additional silicon features".

    "Features are enabled through a license activation process," he wrote. "The SDSi driver provides a per-socket, ioctl interface for applications to perform three main provisioning functions." Those provisioning functions are:

    Continue reading
  • Chip manufacturers are going back to the future for automotive silicon

    Where we're going, we don't need 5nm

    Analysis Cars are gaining momentum as computers on wheels, though chip manufacturers' auto focus isn't on making components using the latest and greatest fabrication nodes.

    Instead, companies that include Taiwan Semiconductor Manufacturing Co and Globalfoundries are turning back the clock and investing billions in factories that use older manufacturing techniques to make chips for vehicles.

    The rapid digitization and electrification of cars has created a giant demand for smaller, more power-efficient auto chips, said Jim McGregor, principal analyst at Tirias Research. He added that cars don't necessarily need the latest manufacturing processes, though, and many are still using analog-based components for various functions.

    Continue reading

Biting the hand that feeds IT © 1998–2021